The Hidden Data Inside Your Files: What Metadata Reveals About You (and How to Strip It)

You attach a photo to a job application. You email a PDF invoice to a client. You upload a CV to a portal. In every one of those files, sitting quietly next to the content you meant to send, is a second layer of information you never typed: where the photo was taken to within a few metres, what phone took it and at what second, the real name on the laptop that made the PDF, the name of the previous client whose invoice you edited into this one.

That second layer is metadata — data about the file rather than data in it. Almost nobody looks at it. It is, quietly, one of the most common ways people leak information they never intended to share.

Three Layers, Not One

"Metadata" gets used as a single word for three quite different things, and knowing which layer you're dealing with tells you whether it survives a copy, an upload, or a conversion.

1. Filesystem metadata — created/modified dates, permissions, the filename itself. This lives in the operating system, not in the file. Copy the file to a USB stick and most of it is rewritten. It also means renaming a file changes nothing about its contents: budget.pdf and holiday-photos.pdf can be byte-for-byte identical.

2. Format-level metadata — structured fields the file format itself defines a place for. EXIF in a JPEG, the document info dictionary and XMP packet in a PDF, the docProps XML inside a .docx. This travels with the file. Email it, upload it, download it again — it's still there.

3. Residual content — not a field at all, but real content still physically present in the file even though you can't see it on screen. Cropped-away image pixels. Text hidden under a black redaction box. Tracked changes and comments in a document you "accepted all" on but never inspected. This is the layer that causes the worst surprises, because there's no "metadata" checkbox anywhere that removes it.

What's Actually in a Photo

EXIF (Exchangeable Image File Format) is the tag block every digital camera and phone writes into a JPEG or HEIC. A typical smartphone photo carries:

The GPS field is the one worth internalising. If location services are enabled for your camera app — the default on most phones — every photo you take is a geotagged record of where you were standing. That's fine for holiday photos in your own library. It's a different matter when you post a picture of your new bike, your desk setup, or your child in a school uniform.

This is not theoretical. In 2012, journalists published a photo of fugitive tech founder John McAfee taken on an iPhone; the EXIF GPS coordinates were intact and revealed he was in Guatemala. It took the internet roughly a day to work it out from the file alone.

The camera serial number matters for a subtler reason: two photos from unrelated accounts that share a body serial were shot by the same physical camera. That's enough to link pseudonymous identities together.

What's Actually in a PDF

A PDF has a document info dictionary and, usually, an XMP metadata block. Common fields:

• Author — very often your real name, pulled from your operating system account. If your computer's user account is your full legal name, that's what lands in every PDF you export.
• Producer and Creator — the software chain, e.g. Microsoft® Word for Microsoft 365 then Adobe PDF Library 24.1. Useful to an attacker for version-specific exploits, and mildly embarrassing when your "professionally designed" brochure says it came out of Word.
• CreationDate / ModDate — with timezone offset, which narrows down roughly where in the world you were.
• Title — frequently the original filename, which is how Invoice-template-COPY-FINAL-v3-usethisone.docx ends up embedded in the polished PDF you sent a customer.

Then there's the residual layer, which is far more serious:

Black rectangles are not redaction. Drawing a filled box over text in a PDF annotator adds a rectangle on top. The text object underneath is untouched — select it, copy it, paste it into a text editor, and it's all there. Real redaction removes the underlying content stream. Government agencies and law firms have repeatedly published "redacted" documents whose hidden text was recovered in minutes with nothing more sophisticated than Ctrl+A, Ctrl+C.

Cropped images keep what you cropped. Crop a photo inside Word or PowerPoint and the application stores the full original and simply displays a window onto it. The cropped-out region — the rest of the room, the other people, the document on the desk — is still in the file until you explicitly run Compress Pictures and tick Delete cropped areas of pictures. Export that document to PDF with the image embedded and the same risk can carry over.

What's Actually in a Word or Excel File

A modern .docx or .xlsx is a ZIP archive — rename it to .zip and open it, and you can read docProps/core.xml and docProps/app.xml in any text editor. Inside:

• Creator and lastModifiedBy — often two different people, which reveals the file was passed between them
• Revision number and TotalTime — how many times it was saved and how many minutes were spent editing it. A cover letter with a total editing time of four minutes tells its own story.
• Company — whatever was typed into Office at install time, sometimes a previous employer
• Comments and tracked changes — invisible if the view is set to "No Markup", entirely present in the file. This is the single most common source of accidental disclosure in professional documents: internal notes like "don't tell them about the delay" shipping to the client because the reviewer's markup was hidden, not removed.

What Strips Metadata — and What Doesn't

This is where most advice gets vague, so here's the practical breakdown.

Reliably removes EXIF:

• Taking a screenshot of the image. The screenshot is a new file made from the pixels on screen. No camera data can survive, because none was involved. Crude, lossy, and effective.
• Re-encoding the image — converting the format or resizing it. The pixels are decoded and written into a brand-new file, and most conversion pipelines simply don't copy the old tag block across. FileConvy's HEIC to JPG converter behaves this way by design: image data and orientation are preserved, detailed EXIF including GPS is not carried into the JPG.

Does not remove anything:

• Renaming the file. Zero effect on contents.
• Zipping it. The file inside is unchanged.
• Cropping in a viewer that supports "revert". If it can undo, the data is still there.
• Emailing it. Attachments are transferred byte-for-byte.

Partially removes — don't rely on it:

• Uploading to a social platform. Facebook, Instagram and X generally re-encode uploads and drop EXIF from the copy other users can download. That protects you from other users — but the platform read the original on the way in, and "the version the public can download is clean" is not the same as "you didn't disclose your location." Messaging apps vary: some strip, some forward the original untouched when you send a file "as a document" rather than as a photo.

The one that matters most: for anything genuinely sensitive, the safe move is not to strip metadata but to rebuild the file from its visible content — convert it to images and back, or re-create it — so that only what you can see on screen exists in the file at all. Converting a PDF to PNG flattens everything: hidden text under a redaction box becomes ordinary pixels, and there is nothing underneath left to recover.

A 60-Second Pre-Send Check

Before you send a file to anyone outside your organisation:

1. Photos — convert or resize the image rather than attaching the camera original. Re-encoding is the cheapest way to drop the GPS tag.
2. PDFs — open Document Properties (Ctrl+D in most readers) and read the Author, Title, and Producer fields. Fix them at the source: if every export says the wrong name, change your Office or OS account display name once and the problem stops recurring.
3. Any redaction — select the text under the black box and try to copy it. If anything pastes, it was never redacted. Flatten the page to an image instead.
4. Office documents — switch the review view to All Markup before sending, not after. Then run File → Info → Check for Issues → Inspect Document.
5. Cropped images inside documents — Compress Pictures → Delete cropped areas of pictures.

What's Coming Next on FileConvy

A few things we're actively building that touch this directly:

• A true AI image upscaler. Today the Image Upscaler enlarges with high-quality smoothing — good for making an image bigger without jagged edges, but it cannot invent detail that was never captured. We're adding a model-based upscaler that reconstructs finer detail, which matters when the only copy you have of a document or photo is a small, already re-encoded one.
• Multi-language OCR. OCR PDF recognises English well today; multi-language recognition is in progress, so scanned documents in Urdu, Arabic, Spanish, Hindi and more become searchable rather than staying flat images.
• The PDF editor leaving beta. Edit PDF is labelled Beta for an honest reason — a PDF is a fixed-layout format, not a document you can retype into. We're working toward genuine in-place text editing rather than a convert-edit-convert round trip.

If there is a tool you keep wishing existed, tell us — the roadmap above is shaped mostly by what people ask for.

Quick FAQ

Does converting an image remove EXIF data? Usually yes. Conversion decodes the pixels and writes a new file, and most pipelines don't copy the old EXIF block into the output — FileConvy's HEIC to JPG conversion, for example, does not carry GPS or camera settings into the JPG. Verify rather than assume: after converting, open the new file's properties and confirm the location field is empty.

Can someone find my home address from a photo I posted? If the photo still has its GPS tag and was taken at home, yes — the coordinates are precise enough to identify a building. Major social platforms strip EXIF from the copy other users download, but photos shared by email, messaging apps in "document" mode, cloud links, or your own website often keep it intact.

Is blacking out text in a PDF the same as redacting it? No, and this is the most dangerous misconception in the whole subject. A black rectangle is drawn on top; the text underneath is still selectable and copyable. To redact for real, remove the content or flatten the page into an image so the text no longer exists as text.

Does renaming or zipping a file remove its metadata? Neither does anything at all. Both leave the file's internal bytes untouched. Only re-encoding, explicit metadata removal, or rebuilding the file from its visible content changes what is inside.

How do I stop my real name appearing as the Author of every PDF I make? Fix it at the source. The Author field is usually copied from your operating system or Office account display name, so changing that once stops the leak for all future exports — far more reliable than editing the field on every individual file.

Related Free Tools

• HEIC to JPG — re-encodes iPhone photos to JPG without carrying GPS and camera EXIF across
• Compress Image — shrink a photo before sending; re-encoding is also the simplest EXIF drop
• PDF to PNG — flatten pages to images so hidden text under redaction boxes stops being text
• Crop Image — crop destructively into a new file, unlike cropping inside Word or PowerPoint
• Why You Should Preview Files Before and After Converting — the habit that catches this class of mistake